v1.3.2

This is a patch release v1.3.2 with no breaking changes.

The following changes have been introduced in this release:

• Use new Helm stable repo URL for Fluent Bit #1079
• Use new Helm stable repo URL for metrics-server #1081
• Add OTLP HTTP port to the OpenTelemetry Collector #1088

v1.3.1

This is a patch release v1.3.1 with no breaking changes.

The following changes have been introduced in this release:

• Add priorityClassName to events statefulset #1006
• Upgrade OpenTelemetry Collector to v0.12 #1013
• Upgrade telegraf-operator to 1.1.6 (official 1.1.5 with helm2 support) #1010
• Override Fluent Bit image to 1.6.0 for Helm chart version 1.3 #1019
• Upgrade fluent-plugin-kubernetes_metadata_filter from 2.4.1 to 2.5.2 #998

v1.3.0

This is a features release v1.3 with no breaking changes.

• With this release we are adding support for app metrics from NginX, Redis and JMX with the help of the Telegraf Operator #893
  Documentation of how to setup the Telegraf can be found here.

• Additionally we migrated Prometheus Operator from the Helm Stable repository to the new Prometheus Community Helm Charts repository #908

  • We still use the prometheus-operator alias in the values.yaml file
  • We will migrate off the prometheus-operator alias into kube-prometheus-stack in the upcoming v2.0 release

• Also we have prepared the performance recommendations document #900

List of main fixes and features:

• Include OpenTelemetry Collector logs by default #960
• Add resource limits for prometheus container #959
• Thanos sidecar has no limits set #862
• Add crio parser to fluent-bit #950
• Introduce cache refresh variation #952
• Fix handling api versions in fluent-plugin-enhance-k8s-metadata #942
• Fix kubeclient v4.9.1 usage with group apis #927
• Do not proxy kubernetes internal traffic #920
• Don't scrape from headless services #895
• Support for Security Context for containers in sumologic-kubernetes-collection #786
• Capture HPA metrics #884
• Expose priority class for resources deployed by k8s collection #732
• Add Source Category customization options for default logs source #749
• Support PodDisruptionBudget for fluentd-logs/metrics/events #742
• Support setting nodeselector for job and pods created in values.yaml #747
• Set limits for job pod used during install #740

Full list of changes is available at https://github.com/SumoLogic/sumologic-kubernetes-collection/milestone/6?closed=1

After installing this release with Helm one may see the following message:

WARNING: File persistence for fluentd is disabled. This might lead to loss of data in case of memory buffer overflow.
We recommend turning this property on for production environments.

We highly recommend to enable file persistence for production environments by setting fluentd.persistence.enabled=true.
This property will be enabled by default in the v2.0 release.

v1.2.3

This is a patch release v1.2.3 with no breaking changes.

The following bugs have been fixed in this release:

• Do not proxy kubernetes internal traffic #920

v1.2.2

This is a patch release v1.2.2 with no breaking changes.

The following bugs have been fixed in this release:

• Retry creating kubeclients in FluentD when error #855
• Relabel pod and service dimensions for non-pod metrics #878
• Drop container=pod label when scraping container network metrics #879

v1.2.1

This is a patch release v1.2.1 with no breaking changes.

The following bugs have been fixed in this release:

• Revert removal of traces config #868
• Use fixed kubeclient version 4.9.0 for gems and docker #870

v1.2.0

This is a features release v1.2 with no breaking changes.

• When installing this release with Helm please make sure to use the --force flag
  The --force is needed because of the labels and annotation changes on statefulsets and deamonsets for #797

• After installing this release with Helm one may see the following message:

WARNING: File persistence for fluentd is disabled. This might lead to loss of data in case of memory buffer overflow.
We recommend turning this property on for production environments.

We highly recommend to enable file persistence for production environments by setting fluentd.persistence.enabled=true.
This property will be enabled by default in the 2.0 release.

• FluentD image has been updated from 1.8.1 to 1.11.1

List of main fixes and features:

• Upgrade fluentd base image, refactor docker file to pull steps into builder image, remove dependencies that are not needed #775
• Ability to inject additional labels to fluentD statefulsets #728
• Helm chart support for setting custom annotations on statefulsets and daemonsets #797
• Expose ability to set additional settings in fluentd forward input plugin #830
• Unify comments in values.yaml - make them start with two hash characters #836
• Set text compress as default option for FluentD file buffer #850

Full list of changes is available at https://github.com/SumoLogic/sumologic-kubernetes-collection/milestone/5?closed=1

v1.1.0

This is a features release v1.1 with no breaking changes. With this release, Helm3 is officially supported.
It contains the following upgrades:

• Support installation behind proxy #692
• Dynamic generation of terraform properties #705
• Enable wal compression in Prometheus by default #719
• Increase fluentd evets sts resource limits #744
• Bump supported k8s versions #737
• Lower the retry_max_interval to 10m for fluentd buffer #768
• Make it optional to collect kubernetes logs, metrics and traces #678
• Allow for some customization of Setup Job #643
• Update falco helm chart #670

And also following changes:

• Logs from Systemd do not preserve originating host in _sourceHost #724
• Control plane metrics #561
• Update fluent-bit helm chart #669
• Update metrics server helm chart #668
• Update prometheus operator helm chart #667
• Add dynamic generation of terraform kubernetes object #675
• List metrics forwarded to the sumologic #629
• Expose Scrape Intervals in values.yaml to make it easier to adjust. #665
• Ensure clusterName has no spaces to prevent issue with Explore rendering. #679
• add collectionMonitoring logic #682
• expose extraEnvVars, extraVolumes, extraVolumeMounts #681
• Use env for prometheus metrics namespace #690
• optional metrics resources #683
• Expose http sources for metrics in values.yaml #672
• Change default retry_timeout to 1h #700
• Handle Collector name with spaces #706
• expose the k8s api version and groups as params #565
• add db for systemd #713
• Otelcol config update with batch size and source processor #715
• Add generator for logs, events and traces sources #708
• Split sumo-k8s.tf on multiple terraform files #721
• set correct source host for systemd logs #725
• Use sumologic.com as prefix for internal labels #734

All the changes can be found here: https://github.com/SumoLogic/sumologic-kubernetes-collection/milestone/4?closed=1

v1.0.0

This is a breaking change release.

It contains the following changes and fixes:

• Falco installation is disabled by default.
• Bumped the helm Falco chart version to v1.1.6 which included a fix to disable the bitcoin/crypto miner rule by default
• The values.yaml file has had several configs moved and renamed to improve usability. Namely, we introduced a new fluentd section into which we moved all of the Fluentd specific configs, while configs for our dependency charts (prometheus-operator, fluent-bit, metrics-server, falco) have not changed.
• The sumologic.kubernetesMeta and sumologic.kubernetesMetaReduce have been removed. The default log format (fluentd.logs.output.logFormat) is fields, which removes the relevant metadata from the JSON body of the logs, making these configs no longer necessary.
• The sumologic.addStream and sumologic.addTime (default values were true) have been removed; the default behavior will remain the same. To preserve the behavior of addStream = false or addTime = false, you can add the following config to the values.yaml file
• Expose the ability to modify the Fluentd configuration
• The Fluentd deployments have been changed to statefulsets to support the use of persistent volumes. This will allow better buffering behavior. They also now include "fluentd" in their names.
• The unified Fluentd statefulsets have been split into set of two different Fluentd's, one for logs and the other one for metrics

For our users who upgrade from the v0.17 release please find the v1.0 upgrade doc.

v0.17.4

This is a patch release. It contains the following fixes:

• add baseImage for Thanos #655 (backport of #650)